486 
Selective Elimination in Staphylea 
errors of random sampling*. In making this test we calculate the frequencies to 
be expected in the classes of a series n if they occurred in the same proportionate 
frequency as in the series m, and determine the chances against this being due 
merely to the errors of random sampling from Elderton's Tablesf. 
Only the critical comparison C — A need be tested. Table XXI gives the 
data. equals 510"8, a value over seven times as large as Elderton has thought it 
TABLE XXI. 
Formula 
Observed 
Frequency 
Calculated 
Frequency 
3 " even " 
1265 
768 
-497 
321-6 
2 " even," 1 " odd " 
928 
1017 
+ 89 
7-8 
1 " even," 2 " odd " 
418 
740 
+ 322 
140-1 
3 "odd" 
93 
179 
+ 86 
41-3 
Totals 
2704 
2704 
0 
510-8 
worth while to table. The chances are thus many millions to one against such a 
divergence as this occurring by the accidents of random sampling. 
Why locules with an odd number of ovules should be inferior to those with an 
even number, or fruits composed in part or altogether of " odd " locules should be 
inferior to those composed chiefly or altogether of " even " locules, our data do not 
permit us to determine. There can be no reasonable doubt of the fact of the 
relationship for our material. 
Problem 5. Is there a difference in the percentage of 2- and 4-merous 
ovaries among the eliminated ovaries and those which develop to maturity ? 
The answer to this question cannot be expected to be very conclusive 
because of the rarity of the anomalies in the series under consideration. The 
available data are given in Table XXXVII. The reason for the frequencies 
for 3-merous fruits being higher in Series C in this Table than in those given 
before is that, to determine more accurately the percentage of anomalous fruits, 
larger samples were examined for this point without counting the numbers 
of ovules. 
For present purposes using only the grand totals, we find the frequencies for 
2-loculed fruits : 
Series A = 2-477 per cent. 
Series B = 2'321 per cent. 
Series C = 1'488 per cent. 
B— A = - -156 per cent. 
C— B = - -833 per cent. 
C— A = - -989 per cent. 
* Pearson, K. : Phil. Mag. Vol. l. pp. 157—175, 1900. 
t Elderton, W. P.: Biome.trika, Vol. i. pp. 155—163, 1901. 
